Synchronized Ultrasonic Vocalizations and Rat Behavior

Ultrasonic vocalisations were recorded at a sampling rate of 384 kHz with an ultrasonic microphone (sensitivity range: 10 to 160 kHz; M500−384, Pettersson Elektronik, Sweden) attached 10 cm above the cage lid and a freeware sound-recording programme (Audacity 2.1.3).¹ Sound events were identified from spectrograms generated in Audacity with fast Fourier transform (Hanning window of size 1024). USVs were defined as discrete sonic events of peak frequency 20−100 kHz and duration 10−150 ms, with a minimum of 20 ms between 2 events. We chose such broad criteria to be inclusive of any ultrasonic events not previously described in the literature. The current classification of 50 kHz USV subtypes sometimes lack information on frequency, we therefore recorded subtype and estimated frequency independently. For USV subtype identification, we used published USV descriptions [following amphetamine injection, as described in Wright et al. (2010) (link)], and our own experience in detecting USVs emitted during rat tickling or play behaviour (Bombail et al., 2019 (link); Hammond et al., 2019 (link)). When further confirmation was required, the recordings were played at 0.05 × speed to listen to the sounds in the human audible range. To visually assess USV frequency, we estimated the projected median value of the total USV trace onto the frequency axis (to the nearest multiple of 5 kHz). Audacity recordings were visually assessed by two researchers naïve to experimental conditions.
Rats were video recorded during the test meals (Sony 12.0 mega pixels HDR-XR-500 Handycam). Recordings were analysed using a freeware video player that allows frame by frame analysis at the resolution of 40 ms (64-bit PotPlayer).² For each rat, the video and USV track timings were synchronised using the sound of a timer audible on both recordings, at the start of each phase. The timing of this event could be assigned to a video frame (at the resolution of 40 ms) and to the USV spectrogram (to the nearest millisecond). This was used to investigate what actions the rats were performing upon USV production. We also used scan sampling at the resolution of 1 s to describe behaviour over the 10 min of anticipation or food consumption phases (600 observations per individual and phase). The ethogram of all the behaviours detected in the recordings is described in Table 1, it follows criteria based on our previous work (Bombail et al., 2019 (link), 2022 (link); Hammond et al., 2019 (link); Champeil-Potokar et al., 2021 (link)).